Generalized Measures of Anticipation and Responsivity in Online Language Processing
This work addresses the need for more expressive measures in psycholinguistics, offering incremental improvements for researchers studying language processing.
The authors tackled the problem of measuring predictive uncertainty in online language processing by generalizing classic information-theoretic measures, resulting in new measures that showed enhanced predictive power for human cloze completion probability and neural amplitudes, and greater complementarity with surprisal in predicting reading times.
We introduce a generalization of classic information-theoretic measures of predictive uncertainty in online language processing, based on the simulation of expected continuations of incremental linguistic contexts. Our framework provides a formal definition of anticipatory and responsive measures, and it equips experimenters with the tools to define new, more expressive measures beyond standard next-symbol entropy and surprisal. While extracting these standard quantities from language models is convenient, we demonstrate that using Monte Carlo simulation to estimate alternative responsive and anticipatory measures pays off empirically: New special cases of our generalized formula exhibit enhanced predictive power compared to surprisal for human cloze completion probability as well as ELAN, LAN, and N400 amplitudes, and greater complementarity with surprisal in predicting reading times.